Resolution criteria
This market resolves to YES if, before the market's specified close date (or by an explicit deadline set by the creator), Anthropic publicly announces or publishes a formal proof (or counterexample) settling the Extended Riemann Hypothesis (ERH) (NOT THE ALREADY RESOLVED Generalized Riemann Hypothesis), and that proof achieves widespread academic consensus or formal verification.
Proof standard: The proof must either:
Be published in a peer-reviewed, reputable mathematical journal or accepted by a major mathematical body (such as the Clay Mathematics Institute or the International Mathematical Union), OR
Be formally verified end-to-end in an interactive theorem prover (e.g., Lean, Isabelle, or Coq) and confirmed by independent researchers.
Attribution: The result must be directly attributed to Anthropic—either authored by Anthropic researchers acting on behalf of the company, or generated primarily by an Anthropic AI system/model.
Resolution sources: Peer-reviewed publications, the Clay Mathematics Institute, or official announcements from Anthropic's research blog accompanied by verification from the broader mathematics community.
Otherwise, or if no verified proof settling the hypothesis is presented before market close, this market resolves to NO.
Note: The creator may specify a precise cutoff date in the market settings if one has not already been established.
Background
The Extended Riemann Hypothesis (ERH) is a generalization of the Riemann Hypothesis to all Dedekind zeta functions of number fields (or more broadly, to certain classes of $L$-functions). While the standard Riemann Hypothesis concerns the zeros of the Riemann zeta function $\zeta(s)$, the ERH asserts that all non-trivial zeros of Dedekind zeta functions also lie on the critical line $\text{Re}(s) = 1/2$. A resolution to the ERH would have profound consequences across analytic number theory, algebraic number theory, and computational complexity (such as deterministic polynomial-time primality tests and bounds on characters).
While AI labs, including Anthropic, have increasingly deployed frontier models toward formal mathematics, theorem proving, and code verification (such as integration with Lean), proving major open conjectures in analytic number theory remains an unsolved challenge for autonomous AI systems.
Market was 75% when I bet; I have it around 12%. Today's Lean proof from Alpöge covers Dirichlet L-functions (GRH), not the Dedekind zeta functions ERH needs. Even that proof isn't independently checked yet, and this needs a verified ERH proof by Dec 31. https://terrytao.wordpress.com/2026/11/20/unforced-euler-singularity-unforced-navier-stokes-singularity-and-the-generalized-riemann-hypothesis/